Incremental encoding

Incremental encoding, also known as front compression, back compression, or front coding, is a type of delta encoding compression algorithm whereby common prefixes or suffixes and their lengths are recorded so that they need not be duplicated. This algorithm is particularly well-suited for compressing sorted data, e.g., a list of words from a dictionary.

For example:

Input

Common prefix

Compressed output

myxa
myxophyta
myxopod
nab
nabbed
nabbing
nabit
nabk
nabob
nacarat
nacelle

no preceding word
'myx'
'myxop'
no common prefix
'nab'
'nabb'
'nab'
'nab'
'nab'
'na'
'nac'

0 myxa
3 ophyta
5 od
0 nab
3 bed
4 ing
3 it
3 k
3 ob
2 carat
3 elle

64 bytes

46 bytes

The encoding used to store the common prefix length itself varies from application to application. Typical techniques are storing the value as a single byte; delta encoding, which stores only the change in the common prefix length; and various universal codes. It may be combined with other general lossless data compression techniques such as entropy encoding and dictionary coders to compress the remaining suffixes.

Applications

Incremental encoding is widely used in information retrieval to compress the lexicons used in search indexes; these list all the words found in all the documents and a pointer for each one to a list of locations. Typically, it compresses these indexes by about 40%.[1]

As one example, incremental encoding is used as a starting point by the GNU locate utility, in an index of filenames and directories. The GNU locate utility further uses bigram encoding to further shorten popular filepath prefixes.

gollark: Probably just that while people like the idea of better-compressed images, it's not very useful for a browser or whatever to implement it if no sites use it, and not very useful for a site to implement it if no browsers support it.

gollark: I'm not really sure.

gollark: No, at least in this field they're frequently made by large well-funded teams, but it just takes ages for support to be implemented anywhere.

gollark: I mean, apart from support, AVIF is not very good in terms of being supported by anything at all, but it's technologically superior.

gollark: Also, JPEG bad AVIF good.

References

Ian H. Witten, Alistair Moffat, Timothy C. Bell. Managing Gigabytes. Second edition. Academic Press. ISBN 1-55860-570-3. Section 4.1: Accessing the lexicon, subsection Front coding, pp.159–161.

This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.

[1] Ian H. Witten, Alistair Moffat, Timothy C. Bell. Managing Gigabytes. Second edition. Academic Press. ISBN 1-55860-570-3. Section 4.1: Accessing the lexicon, subsection Front coding, pp.159–161.